Benchmark · KnowledgeCompetitive

JMMLU

Updated 2025-01-20
Models tested
12
Top score
72.2
DeepSeek R1 Distill Qwen 32B
Median
43.5
min 24.2
Top-5 spread
σ 8.5
wide open

Best score over time · one chart, every benchmark

JMMLU6 MODELS · FRONTIER RUNNING MAX0255075100SCORE ↑Aug 24Oct 24Nov 24Dec 24Jan 25RELEASE DATE →benchgecko.ai/benchmark/jp-jmmlu · frontier
Frontier on JMMLU rose from 63.4 to 72.2 in 1 months · +8.8 points · latest leader DeepSeek R1 Distill Qwen 32B from DeepSeek.
Pink dots = frontier records · 2 totalClick to open model page

Same category · related evaluations